[llvm] [DependenceAnalysis] Extending SIV to handle fusable loops (PR #128782)

[Alireza Torabian via llvm-commits]

================
@@ -0,0 +1,345 @@
+; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py UTC_ARGS: --version 5
+; RUN: opt < %s -disable-output "-passes=print<da>" -aa-pipeline=basic-aa 2>&1 \
+; RUN:   -da-disable-delinearization-checks | FileCheck %s
+
+
+;;  for (long int i = 0; i < n; i++) {
+;;    for (long int j = 0; j < n; j++) {
+;;      for (long int k = 0; k < n; k++) {
+;;        for (long int l = 0; l < n; l++)
+;;          A[i][j][k][l] = i;
+;;      }
+;;      for (long int k = 1; k < n+1; k++) {
+;;        for (long int l = 0; l < n; l++)
+;;          *B++ = A[i + 4][j + 3][k + 2][l + 1];
+
+define void @Fusable0(i64 %n, ptr %A, ptr %B) nounwind uwtable ssp {
+; CHECK-LABEL: 'Fusable0'
+; CHECK-NEXT:  Src: store i64 %i.013, ptr %arrayidx12, align 8 --> Dst: store i64 %i.013, ptr %arrayidx12, align 8
+; CHECK-NEXT:    da analyze - none!
+; CHECK-NEXT:  Src: store i64 %i.013, ptr %arrayidx12, align 8 --> Dst: %0 = load i64, ptr %arrayidx24, align 8
+; CHECK-NEXT:    da analyze - flow [-4 -3]! / assuming 2 fused loop(s): [-4 -3 -3 -1]!
+; CHECK-NEXT:  Src: store i64 %i.013, ptr %arrayidx12, align 8 --> Dst: store i64 %0, ptr %B.addr.34, align 8
+; CHECK-NEXT:    da analyze - confused!
+; CHECK-NEXT:  Src: %0 = load i64, ptr %arrayidx24, align 8 --> Dst: %0 = load i64, ptr %arrayidx24, align 8
+; CHECK-NEXT:    da analyze - none!
+; CHECK-NEXT:  Src: %0 = load i64, ptr %arrayidx24, align 8 --> Dst: store i64 %0, ptr %B.addr.34, align 8
+; CHECK-NEXT:    da analyze - confused!
+; CHECK-NEXT:  Src: store i64 %0, ptr %B.addr.34, align 8 --> Dst: store i64 %0, ptr %B.addr.34, align 8
+; CHECK-NEXT:    da analyze - confused!
+;
+entry:
+  %cmp10 = icmp sgt i64 %n, 0
+  br i1 %cmp10, label %for.cond1.preheader.preheader, label %for.end35
+
+
+for.cond1.preheader.preheader:                    ; preds = %entry
+  br label %for.cond1.preheader
+
+for.cond1.preheader:                              ; preds = %for.cond1.preheader.preheader, %for.inc33
+  %B.addr.014 = phi ptr [ %B.addr.1.lcssa, %for.inc33 ], [ %B, %for.cond1.preheader.preheader ]
+  %i.013 = phi i64 [ %inc34, %for.inc33 ], [ 0, %for.cond1.preheader.preheader ]
+  %cmp28 = icmp sgt i64 %n, 0
+  br i1 %cmp28, label %for.cond4.preheader.preheader, label %for.inc33
+
+for.cond4.preheader.preheader:                    ; preds = %for.cond1.preheader
+  br label %for.cond4.preheader
+
+for.cond4.preheader:                              ; preds = %for.cond4.preheader.preheader, %for.inc30
+  %B.addr.110 = phi ptr [ %B.addr.2.lcssa, %for.inc30 ], [ %B.addr.014, %for.cond4.preheader.preheader ]
+  %j.09 = phi i64 [ %inc31, %for.inc30 ], [ 0, %for.cond4.preheader.preheader ]
+  %cmp53 = icmp sgt i64 %n, 0
+  br i1 %cmp53, label %for.cond7.preheader.preheader, label %for.cond15.loopexit
+
+for.cond7.preheader.preheader:                    ; preds = %for.cond4.preheader
+  br label %for.cond7.preheader
+
+for.cond7.preheader:                              ; preds = %for.cond7.preheader.preheader, %for.inc12
+  %k.07 = phi i64 [ %inc13, %for.inc12 ], [ 0, %for.cond7.preheader.preheader ]
+  %cmp81 = icmp sgt i64 %n, 0
+  br i1 %cmp81, label %for.body9.preheader, label %for.inc12
+
+for.body9.preheader:                              ; preds = %for.cond7.preheader
+  br label %for.body9
+
+for.body9:                                        ; preds = %for.body9.preheader, %for.body9
+  %l.02 = phi i64 [ %inc11, %for.body9 ], [ 0, %for.body9.preheader ]
+  %arrayidx12 = getelementptr inbounds [100 x [100 x [100 x i64]]], ptr %A, i64 %i.013, i64 %j.09, i64 %k.07, i64 %l.02
----------------
1997alireza wrote:

%n is removed.

https://github.com/llvm/llvm-project/pull/128782